By Josh Green
Have you ever thought that the local doctor is a little mean to you? Ever thought that the doctor never listened or had it in for you? What if you could get all of the services that they provided, were impartial and also led to practically no waiting times? There is a vast library of information out there for all your medical needs from the dodgy knee to the abnormal pain in your heart and apps have been created to offer a computer to be your own personal GP! It has been estimated that diagnostic errors from human doctors ranges from 10 to 15 percent and, obviously, the ideal goal is for a system that can reduce this down to 0 percent.
So you must be wondering whether these ‘computer GP’s’ are ready to replace our friendly (or not so friendly!) human doctors? Many people already try to use online resources, developed over the past two decades, to discover what could be wrong with them by self-diagnosis. However, direct comparisons between these self-diagnosis tools and human doctors has not been done scientifically. Researchers at the Harvard Medical School have been on the case and found that, as of yet, self-diagnostic apps or internet-based services lag far behind their human counterparts.
In a study published on October 10 th this year in the JAMA Internal Medicine journal the scientists, in the study, demonstrated that twenty three commonly used symptom-checker apps could not even manage to correctly predict half of the cases doctors were presented with. The study gave 234 medical physicians 45 clinical cases to solve with a diagnosis along with two other possible diagnoses.
At least 20 physicians solved each case accurately. Accounting for both the initial guess and the additional two diagnoses given the doctors vastly outperformed the self-diagnosis tools. The first initial guess made by physicians was correct a whopping 72 percent of the time compared to the 34 percent for digital platforms. Including those further guesses, the success rate rose to 84 percent compared to 51 percent with the digital adversaries. A really interesting factor in the study was that the cases most often solved by the physicians rather than the tools were the most medically severe or the least severe cases. The cases that were ‘more common’ was where the gap between the two was the smallest.
The desire of a fully scientific and exact diagnosis system is ultimately desired but it seems that the gut feelings and wisdom of our doctors dominate flawed self-diagnosis apps. This should come as a relief to the doctors out there, for now. As one of the principal researchers of the study, Ateev Mehrotra, states “Clinical diagnosis is currently as much art as it is science” [Credit: Science Daily] and right now apps cannot replace the ‘art’ of experience, gut feeling and knowhow of the very people on the ground; our doctors?